Rising AI repos
Top 10 rising AI repos, week of Nov 2 (ET)
ET · 10 repos
Data: ET · As of · Published
Total stars are a snapshot at that time. Stars gained come from GitHub’s daily star data, whose day boundaries may not line up with ET midnight.
The short version
morluto/rea gained +25,474 stars and held the top spot for a third week, though its gain slipped from about +31k. T3 Code more than doubled its gain to +14,124, and agent-desktop arrived at +12,431.
Three of the ten are new to this chart (agent-desktop, NVIDIA’s Auto Ontology and OpenMausBot). Multi-agent workspaces took three spots (T3 Code, OpenMausBot and Orca), and agent skills took three more.
-
morluto/rea
+25,474 stars gained. One MCP server that lets your agent take apart native binaries, apps and websites, then explain how a feature works down to the binary.
Why it matters for agent builders: It shows how much one MCP can carry when it wraps heavy local tools, a useful model if you plan to hand desktop tools to an agent.
Source: morluto/rea on GitHub
-
pingdotgg/t3code
+14,124 stars gained. An open control surface for the coding agents already on your machine, with mobile, web and desktop apps that drive Claude Code, Codex, Cursor, Grok Build, OpenCode and others.
Why it matters for agent builders: It runs on subscriptions you already pay for, and the MIT license means you can fork it if the project heads somewhere you don’t like.
Source: pingdotgg/t3code on GitHub
-
vercel-labs/agent-desktop
+12,431 stars gained. A CLI that gives an AI agent computer use on macOS, Windows and Linux by reading each app’s accessibility tree as JSON instead of matching screenshots.
Why it matters for agent builders: The agent acts on element refs, not pixels, which skips the screenshot loop. Its README warns that snapshots and screenshots can hold private desktop content.
Source: vercel-labs/agent-desktop on GitHub
-
cloudflare/security-audit-skill
+10,676 stars gained. A coding-agent skill that runs a multi-phase security audit with separate hunter and verifier agents and writes findings against a JSON schema.
Why it matters for agent builders: Pairing an agent that finds problems with one that tries to disprove them is a pattern any reporting agent could borrow.
Source: cloudflare/security-audit-skill on GitHub
-
NVIDIA/auto-ontology
+9,941 stars gained. NVIDIA’s framework for building a semantic layer over company data, so agents know where a metric lives and what it means, served through MCP, a CLI and an API.
Why it matters for agent builders: An agent that answers data questions is only useful if the number is right, and this checks generated queries against certified definitions. It isn’t taking outside contributions yet.
Source: NVIDIA/auto-ontology on GitHub
-
mattpocock/skills
+9,729 stars gained. A set of small, composable agent skills for everyday engineering work, installable as a plugin for Claude Code, Codex and GitHub Copilot.
Why it matters for agent builders: Skills short enough to read and edit in one sitting are an easy alternative to frameworks that take over your whole workflow.
Source: mattpocock/skills on GitHub
-
deepseek-ai/deepseek-harness
+8,421 stars gained. DeepSeek’s open-source agent harness (dsh), built so that everything is a plugin. It’s still a developer preview with breaking changes expected.
Why it matters for agent builders: A model lab’s own harness hints at which plugin points it thinks an agent needs, though the APIs may keep moving while it’s in preview.
Source: deepseek-ai/deepseek-harness on GitHub
-
milind-soni/OpenMausBot
+8,068 stars gained. A local-first desktop chat app where each bot is a Claude, Codex or Grok agent running on your own machine, with its own personality, model and connected apps.
Why it matters for agent builders: It runs on the CLIs and logins you already have, with no new account or proxy in the middle. It’s an open-source take on hosted bot-team apps.
Source: milind-soni/OpenMausBot on GitHub
-
stablyai/orca
+6,930 stars gained. A desktop and mobile workspace that runs a fleet of coding agents side by side, each in its own git worktree, on your own subscriptions.
Why it matters for agent builders: Sending one prompt to several agents and keeping the best result is turning into a habit, and this is one way to keep track of it.
Source: stablyai/orca on GitHub
-
pbakaus/impeccable
+6,822 stars gained. A design vocabulary for coding agents. Steer with polish and bolder, review with critique and audit, and catch AI-looking defaults before they ship.
Why it matters for agent builders: Its gain cooled from about +10.8k last week, but a shared vocabulary is still a low-effort way to push agents off the same cream-and-purple default.
Source: pbakaus/impeccable on GitHub
What to watch next
- rea’s weekly gain has eased for two straight weeks, from about +60k to +31k and now +25k. Next chart shows whether it keeps the top spot.
- pingdotgg/t3code more than doubled its gain in its second week on the chart. See whether remote control of agents you already pay for keeps pulling people in.
- vercel-labs/agent-desktop went from a handful of stars a day in late October to +12,431 this period. Watch whether accessibility-tree computer use holds interest after launch week.